Mneme HQ deploys inside the developer platforms engineering organizations already run — Microsoft Azure & GitHub Enterprise, AWS, Google Cloud, and self-hosted environments. It does not provision cloud resources or manage IAM. It governs the AI coding agents that operate inside the platform, on every CI run and every editor session.
Platforms

Works within enterprise engineering environments

Deploy Mneme HQ across your existing developer platform stack. Policy enforcement for AI coding agents in CI/CD, runtime, and IDE surfaces — portable across clouds, agents, and editors.

Platforms Mneme runs inside

Four enterprise developer platform contexts where AI coding agents actually run. Mneme is the governance layer above the runtime in each.

Microsoft · Azure AI & GitHub EnterpriseEnterprise

Where Mneme fits: as a GitHub Actions governance check on PR diffs and as a context provider for Copilot, Claude Code, and other agents operating in Azure DevOps and GitHub Enterprise.

Designed to support
  • GitHub Actions — mneme check --mode strict as a required check on every PR.
  • GitHub Copilot governance — the decision corpus is the source of truth Copilot's suggestions get gated against in CI.
  • Azure DevOps pipelines — the same CLI runs on Azure-hosted runners.
  • Azure AI Foundry / OpenAI deployments — model-agnostic, so swapping the Azure-hosted model does not affect governance.
AWS · Developer EnvironmentsEnterprise

Where Mneme fits: inside CodeCatalyst pipelines, on self-hosted Bedrock-backed agent workflows, and on EC2 / ECS-hosted CI runners.

Designed to support
  • CodeCatalyst CI workflows — same governance gate, AWS-hosted runner.
  • Bedrock-hosted coding agents — OpenAI-compatible inference; governance applies regardless of the model behind the endpoint.
  • Self-hosted GitHub Actions runners on AWS — for teams that want repo-platform on GitHub Enterprise but compute inside their AWS account.
  • Lambda / Step Functions agent pipelines — the CLI is a small Python install; can run inside short-lived compute.
Google Cloud · AI EngineeringEnterprise

Where Mneme fits: as a Cloud Build step, alongside Gemini Code Assist in the editor, and on Vertex-hosted agent runtimes.

Designed to support
  • Cloud Build pipelines — the governance gate runs as a step on every PR/merge.
  • Gemini Code Assist — same decision corpus, same enforcement surface.
  • Vertex AI Agent Builder / agent runtimes — context provider for custom agents operating on Vertex.
  • Cloud Workstations — local-mode mneme check --mode warn in the developer's dev environment.
Self-Hosted / Air-GappedOn-prem

Where Mneme fits: on any internal developer platform that runs git, CI, and an inference endpoint. No outbound calls required.

Designed to support
  • Self-hosted inference — vLLM, Ollama, Triton, on-prem Anthropic or OpenAI-compatible gateways.
  • Self-hosted GitLab CI / Jenkins / Bamboo — CLI is a single Python entry point; runs on any runner.
  • Internal developer platforms (IDPs) — Backstage, Cortex, OpsLevel; the decision corpus is a tracked file the IDP can surface.
  • No telemetry, no hosted control plane — the entire system runs inside the network.

Cross-cutting capabilities. The same governance surface applies on every platform above: CI/CD execution gates, agent and runtime context injection, enterprise developer workflow policy enforcement, and governance portability across heterogeneous coding agents. The decision corpus is a tracked file in the repo; whichever platform runs the CI and whichever agent generated the code, the rules are the same.

Scope · what Mneme is and is not

What Mneme does on these platforms

  • Governs AI coding agents inside the developer workflow
  • Enforces architectural decisions at the file-write seam and in CI
  • Provides one decision corpus across multiple agents and clouds
  • Surfaces violations before merge, regardless of which model or agent generated the code

What Mneme is not

  • Not cloud infrastructure provisioning (no Terraform-replacement, no Kubernetes setup)
  • Not identity / access / posture management (no IAM, no CSPM, no secrets management)
  • Not a security or compliance scanner (no SAST, no SCA replacement)
  • Not a managed service — it's an open-source CLI plus a JSON corpus the team owns
Core positioning
The platform runs the agents.
Mneme governs the agents.

The cloud is downstream. The agent is downstream. The decision corpus is the constant your engineering org owns — portable across Azure, AWS, Google Cloud, and the air-gapped environment behind the firewall.

Frequently asked

Is Mneme an infrastructure governance tool?
No. Mneme governs AI coding agents inside an engineering organization's existing developer platform. It does not provision cloud resources, manage IAM, or enforce infrastructure policy. It sits above the runtime and inside the developer workflow, regardless of which cloud the workflow lives on. See heterogeneous-agent governance for the structural argument.
Will this fit into our Azure / GitHub Enterprise / Copilot stack?
Yes. Mneme runs as a GitHub Actions check on PR diffs and as a context provider for Copilot, Claude Code, Cursor, and other agents operating in the Azure DevOps and GitHub Enterprise workflow. The decision corpus is a tracked file in the repo, so it inherits whatever access controls the platform already enforces. See the GitHub Actions integration.
Does Mneme require a hosted control plane?
No. Mneme is an MIT-licensed CLI plus a structured decision corpus. It runs on the laptop, on a CI runner, or on a self-hosted runner inside an air-gapped environment. There is no hosted service to subscribe to and no telemetry that has to leave the network.
How does Mneme compare to platform-native AI tooling?
Platform-native AI tooling (Copilot, Bedrock-hosted agents, Gemini Code Assist) is the agent. Mneme is the governance layer above the agent. It is designed to operate alongside any of them, providing one decision corpus and one enforcement surface even when an organization runs multiple AI coding tools across multiple clouds. See Works With for the full surface.
What about multi-cloud organizations?
The decision corpus is repo-local. A multi-cloud org running GitHub Enterprise repos with CI on AWS and dev environments on Google Cloud Workstations gets the same governance surface across all three, with no per-cloud encoding of the rules.

Get started

Source GitHub → MIT-licensed. Works on every platform listed above. CI wiring GitHub Actions integration → End-to-end PR-diff governance using mneme check. Reference CLI → Commands and CI patterns — same surface across platforms. Compatibility Works with → The models, agents, and frameworks Mneme operates alongside.